Dear @Jundong.Wang,
I’m not familiar with all of your MATLAB calculations, but are you tying the frequency step to the simulation length? You could try comparing your calculations to the simple Jason_PSD.m script shared on this forum before: User defined spectrum from routine UserWaveSpctrm in HydroDYn - #7 by Jason.Jonkman.
Like @Riad.Elhamoud, I’d say that the tower fore-aft mode is often heavily damped aerodynamically, making its natural frequencies difficult to spot. What you are more likely to see in the tower fore-aft signal is the rotor speed (1P) and its harmonics (3P, 6P, etc.) as the rotor spins through full-field turbulence.
Best regards,